Activation of the microglial neurotoxic response by components of the senile plaque plays a critical role in the pathophysiology of Alzheimer's disease (AD). Microglia induce neurodegeneration primarily by secreting nitric oxide (NO), tumor necrosis factor-α (TNFα), and hydrogen peroxide. Central to the activation of microglia is the membrane receptor CD40, which is the target of costimulators such as interferon-γ (IFNγ). Chromogranin A (CGA) is a recently identified endogenous component of the neurodegenerative plaques of AD and Parkinson's disease. CGA stimulates microglial secretion of NO and TNFα, resulting in both neuronal and microglial apoptosis. Using electrochemical recording from primary rat microglial cells in culture, we have shown in the present study that CGA alone induces a fast-initiating oxidative burst in microglia. We compared the potency of CGA with that of β-amyloid (βΑ) under identical conditions and found that CGA induces 5–7 times greater NO and TNFα secretion. Coapplication of CGA with βΑ or with IFNγ resulted in a synergistic effect on NO and TNFα secretion. CD40 expression was induced by CGA and was further increased when βΑ or IFNγ was added in combination. Tyrphostin A1 (TyrA1), which inhibits the CD40 cascade, exerted a dose-dependent inhibition of the CGA effect alone and in combination with IFNγ and βΑ. Furthermore, CGA-induced mitochondrial depolarization, which precedes microglial apoptosis, was fully blocked in the presence of TyrA1. Our results demonstrate the involvement of CGA with other components of the senile plaque and raise the possibility that a narrowly acting agent such as TyrA1 attenuates plaque formation.
- Alzheimer's disease
- oxidative burst
- nitric oxide
- tyrphostin A1
activation of microglia, macrophages that reside in brain, plays a crucial role in the pathophysiology of neurodegenerative diseases such as Parkinson's disease, amyotrophic lateral sclerosis, and, in particular, Alzheimer's disease (AD) (19). Senile plaques, the histological hallmark of AD, represent a site of inflammation where immune cells, activated by β-amyloid (βΑ) and other plaque components, produce cytotoxic agents that result in neurodegeneration. The immune effectors of the plaque are microglia, which have been shown to play a dual role in the progression of the disease (40). While activated microglia in the senile plaque secrete neurotoxic agents, they can also act as phagocytic cells, removing amyloid fibrils and retarding further neuronal degradation. Agents that influence microglial response to βΑ were shown to be important modifiers of AD progression. These mediators can be divided into protoxic mediators, such as interferon-γ (IFNγ), and prophagocytic mediators, such as tumor growth factor-β (TGFβ) (1, 41). Such interactions exemplify the dynamic nature of the senile plaque and emphasize the need to explore the complex interaction of the multiple components that make up the plaque.
βA is thought to be the principal component of senile plaque, and its secretion by neurons is thought to initiate plaque formation (14). Exposure of microglia to βΑ elicits a neurotoxic response in which nitric oxide (NO) and tumor necrosis factor-α (TNFα) released by microglia are the key effectors (8). While βΑ alone is a moderate inducer of such a response, its effect is subject to synergistic amplification in the presence of cytokines, such as IFNγ (23). The mechanism underlying this effect was shown to be the induction of expression of a tyrosine kinase receptor protein, CD40 (25). CD40 signal transduction is central to a number of chronic inflammatory diseases, including rheumatoid arthritis, systemic lupus erythematosus, and multiple sclerosis (35). Using a CD40 ligand-deficient animal, Tan et al. (33) demonstrated that βΑ activation of microglia is CD40 dependent. In addition, agents that attenuate the synergistic neurotoxic effect of IFNγ, such as TGFβ1 and interleukin-4, were shown to downregulate the expression of CD40 in microglia (23, 39).
A recently described component of the senile plaque that has gained attention in the past few years is the neurosecretory protein chromogranin A (CGA). A number of histological studies have demonstrated its existence in the majority of human senile plaques and have correlated this finding with the preclinical stage of AD (3, 10, 21, 29, 42). A key step in the progression of AD and other neurodegenerative diseases is the production of reactive oxygen species (ROS) by microglia (20, 27, 30). This is illustrated by the reduction in neurodegeneration in patients with AD and in animal models involving antioxidant drug treatment (20). In the present study, we show for the first time that CGA quickly induces a ROS efflux from microglia. In addition to ROS, CGA strongly stimulates microglia to produce NO and TNFα, which cause neuronal cell death in mixed microglia-neuron cultures treated with CGA (4–6, 34, 37). We demonstrate that CGA acts synergistically with either βΑ or IFNγ to stimulate microglia and that this effect can be fully prevented by a selective tyrosine kinase inhibitor, tyrphostin A1 (TyrA1) (12). Furthermore, we demonstrate that IFNγ and βΑ can act together with CGA to strongly upregulate CD40 expression.
In addition to eliciting a microglial neurotoxic phenotype, CGA induces microglial mitochondrial depolarization and, subsequently, apoptosis (17, 18). Reduction in microglial mass and accumulation of microglial debris is thought to inhibit amyloid clearance by phagocytosis and enhance recruitment of additional immune effectors to the site. Therefore, CGA might both enhance the microglial neurotoxic effect and diminish microglial phagocytic activity in senile plaques. These findings implicate CGA as an important stimulator of senile plaque development and stress the need to evaluate its comparative potency and relevance to other plaque components, in particular βΑ.
MATERIALS AND METHODS
CGA was purified from bovine adrenal medullary chromaffin cells as described previously (43). Other material was purchased from the following vendors: TyrA1, Alexis Biochemicals (San Diego, CA); IFNγ, Chemicon International (Temecula, CA); βΑ, Bachem (Torrance, CA); and catalase, Sigma-Aldrich (St. Louis, MO).
Microglial cells were isolated from newborn rats as described previously in detail (13). Briefly, brains were cleared from meninges and blood vessels and then minced with a razor blade. Cells were dissociated by performing trypsinization for 10 min (0.25% trypsin EDTA; Invitrogen, Carlsbad, CA) in phosphate buffer with DNase at 37°C. The dissociated cells were washed twice in PBS supplemented with 10% fetal calf serum (FCS) and transferred to 75-cm2 flasks (Costar, Corning, NY) at a density of 2 × 107 cells/flask. Cells were maintained in 5% CO2 at 37°C in DMEM supplemented with 0.37% NaHCO3 and 10% heat-treated FCS. Medium was changed every 3 days. After 6 days, cells growing on the top of the confluent cell layer were removed by shaking and then plated onto glass coverslips (Nunc Intermed, Naperville, IL) at a concentration of 3 × 106 cells/ml. To remove nonadherent cells and any remaining floating debris, each coverslip was gently washed after 20–30 min. This procedure was shown to result in cultures containing 97% microglial cells (31).
Measurement of hydrogen peroxide generation.
Hydrogen peroxide release from microglial cells was measured electrochemically using a self-referencing sensor with a platinum-tipped microelectrode coated with cellulose acetate as described previously (36). H2O2 was oxidized at +0.6 V against a Ag-AgCl reference electrode. The H2O2 concentration difference was measured at two positions 10 μm apart and orthogonal to the cell surface. Measurements were performed at 10-μm steps away from the cell surface. Cells were kept at 37°C throughout the experiment. The H2O2 flux was calculated according to the modified Fick equation: where flux (J) is equal to the product of the diffusion coefficient (D) of H2O2 and the concentration difference of H2O2 (dI·S) over a known distance (dx).
NO production by microglial cells.
NO in cell culture supernatants was measured spectrochemically using the Griess reagent (Sigma) with sodium nitrite as the standard and a NanoDrop spectrophotometer (NanoDrop, Wilmington, DE) (11). Data were corrected for the background levels of nitrite in the cell-free medium.
TNFα was quantified in cell culture supernatants using an enzyme-linked immunosorbent assay kit (Pierce Biotechnology, Rockford, IL) with a solid phase-bound antibody to capture the rat TNFα from standards and samples. The reporter antibody was horseradish peroxidase-conjugated anti-rat TNFα. Bound TNFα was visualized with hydrogen peroxide and tetramethylbenzidine and read at 450 nm. Results were converted to nanograms per milliliter of TNFα using a standard curve.
Western blot analysis.
Gradient 4–15% gels were obtained from Bio-Rad (Hercules, CA). Anti-CD40, Abcam, was used at a dilution of 1:1,000, and secondary HRP-conjugated mouse anti-rabbit antibody (Invitrogen) was used at a 1:5,000 dilution. As a control, we used an antibody against the abundant mitochondrial protein ABCB10 (Research Genetics, Huntsville, AL). Bands were visualized using LumilightPLUS chemiluminescence (Roche, Indianapolis, IN).
Mitochondrial membrane potential.
The relative mitochondrial membrane potential (ΔΨm) was recorded using the ratiometric voltage-sensitive dye 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1; Molecular Probes, Eugene, OR). At low ΔΨm, JC-1 exists primarily as green fluorescent monomers (527-nm emission) and at highly negative ΔΨm as orange/red fluorescent aggregates (590-nm emission). A Zeiss LSM 510 META was used for confocal microscopic imaging with a 488-nm excitation laser. Cells were incubated in 5 μM JC-1 for 15 min at 37°C to load the dye and then washed three times with 37°C culture medium.
The dose-response curve of CGA and βΑ was fitted to the following hyperbolic function: where R is the level of secretion (NO or TNFα) that was elicited by a concentration, C, of a given activator (βΑ or CGA). R0 is the baseline release level elicited when no substance is added (C = 0). Rmax reflects the maximal level of secretion. AC50 is the concentration needed to elicit 50% of Rmax. The exponent n reflects the steepness of the dose-response curve. This equation was also used to fit the dose-response effect of TyrA1. In this case, Rmax reflects the maximal inhibitory effect on NO or TNFα release, where C and IC50 (rather than AC50) denote the concentration of TyrA1 used and the concentration needed to elicit 50% of the maximal inhibitory effect. All fitting procedures and statistical tests were conducted using KaleidaGraph software (Synergy Software, Reading, PA).
CGA induces an oxidative burst in primary microglia.
CGA-induced H2O2 secretion from microglia was monitored using electrochemical detection with a microelectrode and electrode self-referencing as described previously (36). The self-referencing approach is based on electrochemical detection of a gradient of H2O2 that surrounds cells in culture. This method is advantageous for three reasons. 1) It enables the noninvasive determination of the H2O2 gradient surrounding single microglia cells with high spatial and temporal resolution. 2) It is not influenced by ROS generated in the solution by direct release from proteins and protein aggregates as reported previously for βΑ (32). 3) It is not sensitive to NO, another free radical that is released by these cells (see below).
A platinum-tipped microelectrode was placed at known distances from the surface of microglial cells to measure the H2O2 gradient generated in the cell's vicinity along axes perpendicular to the cell surface. Figure 1 shows a summary of these gradient measurements. In the presence of medium alone, no H2O2 gradient could be detected. Addition of CGA to a final concentration of 15 nM generated a stable gradient of H2O2 around the cell. The H2O2 was most concentrated near the cell surface and decreased to nearly immeasurable levels at an average distance of 25.7 μm (3.7 μm ± SD) from the cell surface. H2O2 release could be observed as early as 4 min after exposure of the cells to CGA. Washing out the CGA by perfusion of the dish with CGA-free medium slightly reduced the magnitude of the flux but did not bring it to baseline. The signal could be fully abolished by perfusing the cells with medium containing a combination of CGA and catalase, which verified that the probe was measuring H2O2. Subsequent washing out of the catalase and CGA unmasked the ongoing oxidative burst because catalase did not enter the cells in significant amounts. Treating microglia with the carrier solution alone containing heat-treated CGA did not induce any production of H2O2 (data not shown). In four experiments for each condition, we found that 1) the average response time for detecting a measurable change (5 fA) was 6.5 min (range, 4–12 min) and 2) after addition of CGA, the duration of activity of the oxidative burst exceeded our 30-min measurement window.
CGA potency and synergism with βΑ.
The secretion of NO and TNFα were determined in response to increasing concentrations of CGA and βΑ. Figure 2 shows the average ± SE dose-response curves measured for NO and TNFα. Our results demonstrate that CGA is a more potent stimulator of microglial neurotoxic response than is βΑ. CGA at 10−3 the concentration of βΑ elicited greater microglial secretion of NO and TNFα. The stimulant concentrations capable of eliciting 50% of the maximal secretory response (activation concentration, AC50) of NO were 16.8 nM for CGA and 22.3 μM for βΑ. The AC50 values for CGA- and βΑ-induced TNFα release were 13.2 nM and 9.1 μM, respectively. Exposure of microglia to increasing concentrations of heat-treated CGA did not elicit a significant change in the secretion of NO or TNFα (Fig. 2). As CGA and βΑ coexist in the senile plaque, we tested the combined effect of the two on the release of NO and TNFα from microglia. The agents were added at concentrations that independently elicited low levels of microglial NO and TNFα secretions (Fig. 2). Treatment of the microglial cells with the combination of CGA and βΑ resulted in NO and TNFα release that was significantly greater than the sum of the effects of either agent alone. Such a synergistic response indicates that both CGA and βΑ target related induction pathways.
Interaction with IFNγ: a key component of the senile plaque.
IFNγ is found in senile plaques and has been shown to potentiate βΑ-induced microglial activation (1). We tested whether IFNγ also potentiates CGA-induced microglial activation by determining the dose-response curves of NO and TNFα release after incubation of microglia with CGA in the presence and absence of IFNγ (Fig. 3). By itself, this concentration of IFNγ elicited a small but significant release of NO (4.3 ± 1.5 μM/24 h) and TNFα (0.45 ± 0.2 ng/ml/24 h) compared with the untreated cells (NO, 1.8 ± 1.3 μM/24 h; TNFα, 0.16 ± 0.3 ng/ml/24 h). Addition of 150 U/ml IFNγ increased the potency of CGA-induced microglial activation, shifting leftward the dose-response curves for NO and TNFα secretion. In the presence of IFNγ (150 U/ml), the AC50 value for NO release was reduced from 16.8 to 6.6 nM, and the AC50 value for TNFα release was reduced from 13.2 to 6.3 nM. The effect of the synergistic interaction of CGA and IFNγ on NO and TNFα secretion was statistically significant (P < 0.001) at all CGA concentrations >4 nM.
Induction of CD40 during microglial activation.
The linked activity of CGA with βΑ and with IFNγ suggests a possible role for CD40 in CGA-induced microglial activation. To test for this possibility, we determined the effect of CGA alone and in combination with either IFNγ or βΑ on CD40 expression. Treatment of microglia with 7.5 nM CGA resulted in the induction of CD40 detected by Western blot analysis (Fig. 4A). Expression of CD40 was further upregulated when CGA was supplemented with IFNγ at 150 U/ml. Similarly, the combination of CGA and βΑ resulted in an even further increase in CD40 expression compared with each agent alone (Fig. 4B). In this experiment, we used CGA at concentrations that elicited approximately a half-maximal response (15 nM).
Tyrphostin A1 effect on microglial activation.
TyrA1 has been shown to inhibit the CD40-induced tyrosine kinase cascade (12). We therefore investigated the effect of TyrA1 on the CGA-induced microglial secretion of NO and TNFα (Fig. 5A). TyrA1 exhibited a dose-dependent inhibition of CGA (15 nM)-induced activity, reducing the release of NO and TNFα secretion with IC50 values of 4.8 μM for NO and 6.6 μM for TNFα. In addition, TyrA1 inhibited the net effect of IFNγ (150 U/ml), βΑ (18 μM), and CGA (15 nM) with IC50 values of 4 μM for NO and 3.1 μM for TNFα.
CGA has been shown to induce microglial apoptosis with an early phase of moderate mitochondrial depolarization (18). To test whether the downstream cascade involving CD40 is also responsible for microglial apoptosis, we monitored ΔΨm in microglia exposed to CGA alone or in combination with TyrA1. We used the ratiometric membrane potential-sensitive dye JC-1 and confocal microscopy. After 24 h in the presence of CGA (15 nM), an apparent mitochondrial depolarization could be seen (Fig. 5B, middle) compared with the control state (Fig. 5B, top). However, when added in combination with 17 μM of TyrA1, CGA (15 nM) had no effect on ΔΨm.
Western blot analysis of the expression of CD40 after treatment with CGA in combination with TyrA1 showed no effect of the latter on CD40 protein level (Fig. 5C). This finding is in agreement with those of previous studies in showing that TyrA1 inhibits the downstream signal transduction cascade of CD40 rather than influencing its expression (12).
CGA is found in pathological structures of a number of neurodegenerative diseases, including AD, Parkinson's disease, and Pick's disease (42). Previous studies demonstrated its capacity to activate microglia and induce neuronal as well as microglial damage (6, 34, 37). In this study, we have demonstrated for the first time that CGA can quickly activate microglia to produce ROS intermediates. We compared the long-term effect of CGA and βΑ on microglia and demonstrated that, in culture, CGA is a more potent activator than βΑ. Furthermore, we have shown that CGA can synergize with βΑ and IFNγ to augment the production of NO and TNFα. The microglial response to the combination of CGA and βΑ includes the upregulation of CD40 protein expression. The inhibition of the CD40 tyrosine kinase cascade using TyrA1 completely abolished CGA-induced NO, TNFα secretion, and subsequent mitochondrial depolarization.
A number of studies support an important role for ROS in the pathophysiology of AD and demonstrate beneficial effects of antioxidants on the progression of the disease (30). Neurons in the vicinity of the senile plaque show increased levels of oxidized proteins, lipids, and DNA. As the prime effectors in the brain inflammatory response, microglia are thought to be the major contributors of ROS, including NO and O2− (7, 28). It is unclear whether any of the senile plaque components are direct stimulants of microglial oxidative bursting, during which they release hydrogen peroxide. While some investigators have demonstrated that βΑ may potentiate the effect of oxidative burst inducers such as phorbol 12-myristate 13-acetate (PMA) (22, 38), others have found that βΑ is a direct inducer of H2O2 production in a variety of immune cells (9). However, it is important to note that the authors of these latter reports used ROS-sensitive fluorescent dyes that do not differentiate between NO and O2−. In addition, Tabner et al. (32) showed that βΑ can readily liberate detectable amounts of hydroxyl radicals in culture media with no cells, which could account for the conflicting reports. The technique we used detects the gradient of specifically cell-derived H2O2. It is not influenced by H2O2, which is not cell derived or derived from other ROS such as NO. Our results indicate that CGA activates the microglial production of ROS within minutes, while heat-treated CGA had no effect. This is the most immediate effect reported for CGA in microglia. Although the heat-induced structural changes in CGA that are responsible for the loss of its effect in microglia are not yet clear, boiling has been shown to change the conformation of CGA (43). Compared with the H2O2 gradient measured around microglial cells stimulated with the robust stimulator PMA (130 nM), CGA (15 nM) elicited a flux ∼40% as large (36).
The potential relevance of CGA to the pathophysiology of AD has been the subject of several investigations. The ability of CGA to elicit microglial activation and produce NO and TNFα suggests that CGA may play a role similar to that of βΑ and act in concert with βΑ to induce the production of these neurotoxic agents.
Variations in culture conditions and stimulation protocols that were used by researchers at different laboratories make it difficult to assess the significance of the effect of CGA as a newly characterized factor. In the present study, we used identical experimental conditions to compare the capacity of CGA with βΑ to induce NO and TNFα secretion in microglia. The concentrations of CGA and βΑ used in this study are comparable to those described before (6, 33). We found that the maximal CGA effect on NO production is 6–7 times that of βΑ and that its maximal effect on TNFα secretion is 5–6 times that of βΑ. Moreover, CGA half-maximal activity (AC50) was reached at 1,000-fold lower concentrations compared with βΑ. Since TNFα was shown to stimulate NO production, it is possible that the observed enhancement of NO production is indirectly mediated through the potentiation of TNFα secretion (8). Kingham et al. (17) examined ΔΨm and caspase activity after exposure of microglia to 10 nM CGA. Under these conditions, the earliest mitochondrial depolarization occurred after 24 h and was upstream of caspase 1 activation (18). The mechanism is thought to involve downregulation of mitochondrial oxidative phosphorylation directly or indirectly by NO (18, 26). Thus the higher capacity of CGA to induce NO production may explain why, as opposed to βΑ, CGA is an inducer of microglial apoptosis.
We show further that CGA and βΑ act together to induce considerable microglial activation. The production of both TNFα and NO production is synergistically enhanced by the coapplication of the two agents. This finding is of importance to the pathophysiology of the early stage of a developing senile plaque, in which CGA is more commonly found (10). In this case, CGA may potentiate the neurotoxic effect of as yet nonpathological concentrations of βΑ. Potentiation of inflammatory mediators is a common feature of cytokine receptors that share similar downstream elements. Of relevance to AD is the remarkable effect of IFNγ on βΑ-induced microglial activation (1). IFNγ acts by inducing CD40, a key initiator of the neuroinflammatory cascade in AD (24, 25). When added in combination with CGA, IFNγ significantly shifted the AC50 of CGA to a lower concentration. This effect on NO production was more pronounced, possibly because of the combined effect of IFNγ and the TNFα secreted by the stimulated cells. Induction of CD40 expression is thought to underlie the synergistic effect of IFNγ with βΑ. This suggests that CGA may itself be an inducer of microglial CD40 expression and/or may act through the upregulation of CD40 cascade.
Western blot analysis of CGA-treated microglia confirmed that CD40 can be induced by CGA alone. CD40 induction by IFNγ was shown to involve the activation of NF-κB, which binds the CD40 promoter, leading to increased CD40 transcription (2, 25). IFNγ activated NF-κB by stimulating the secretion of TNFα, resulting in autocrine activation of NF-κB. Our results indicate that CD40 can be induced to a moderate level in the absence of IFNγ directly by CGA or βΑ. In the case of CGA, we show that CD40 induction does not require TNFα production, because TyrA1, which completely abolished TNFα secretion, did not affect CD40 induction. This finding also is supported by the absence of correlation between the capacity of TNFα production and CD40 expression exhibited by CGA- or βΑ-treated cells. A possible IFNγ/TNFα bypassing mechanism is identified in the work of Kang et al. (16), who showed that microglial NF-κB can be activated by ROS. Since CGA-induced ROS generation is rapid, it is likely to be independent of TNFα synthesis and secretion and thus through TNFα-independent activation of NF-κB.
Previous investigations have shown that in microglia, the initiation of TNFα and NO secretion requires both the ligation of CD40 receptor and the activation of CCAAT/enhancer-binding protein-β by IFNγ (15). Clearly, as opposed to βΑ, CGA can induce a substantial NO release without IFNγ, supporting again the possibility that CGA might interact with downstream components of the IFNγ signal transduction pathway.
Although used largely as an inactive tyrphostin, TyrA1 was recently shown to be a fairly specific inhibitor of CD40 cascade in macrophages, inhibiting CD40 ligand-induced NF-κB translocation to the nucleus (12). Application of TyrA1 completely inhibited the combined effect of βΑ, CGA, and IFNγ on NO and TNFα secretion. This finding emphasizes the overall theme suggesting the use of a common pathway for all three stimulators. Coapplication of TyrA1 resulted in restored ΔΨm. This effect of TyrA1 can be explained by the TyrA1-induced reduction in NO production.
Microglial function in AD has two faces, and the CD40 pathway plays a role in directing the cellular response to stimulants toward beneficial or detrimental outcomes. Studies by Wyss-Coray et al. (41) showed that by reducing CD40 expression, TGFβ1 promotes amyloid clearance and reduces plaque formation. Conversely, IFNγ, which induces CD40, is a strong potentiator of the neurotoxic response induced by βΑ. In this context, our results place CGA together with IFNγ as a CD40 inducer, directing the microglial response toward a neurotoxic pathway. Our results demonstrate the linkage of CGA with other components of the plaque and raise the possibility that a selective inhibitor such as TyrA1 can attenuate plaque formation, both by inhibiting secretion of neurotoxins and by restoring microglial viability. As such, TyrA1 might serve as a pharmaceutical candidate drug targeting the inflammatory components of AD.
G. Twig was supported by a fellowship from the Grass Foundation. S. Graf was supported by a fellowship from the Neal W. Cornell Endowed Research Fund. This project was funded by the Grass Foundation and by National Institutes of Health Grant P41RR001395-21.
We thank Sarah Haigh for help with protein analysis. We thank Drs. Dani Dagan, Barbara Ehrlich, Jeffrey Laskin, Steve Zotolli, Richard Shader, Ayse Dosemeci, and Gal Yaniv for their critical review of the manuscript. We thank the fellows, trustees, and directors of the Grass Foundation of 2001 for valuable discussions throughout the time this work was conducted.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society